WP6 - Short term uncertainty-enabled forecasts for local air quality
Traditionally, air quality measurement networks have been used to provide assessment of the health or environmental impacts of airborne pollutants. Almost every country in Europe has its own on-line and near-real time monitoring service for air quality. Monitoring networks and on-line services are increasingly being complemented by modelling tools, particularly when forecasting air quality. Several air quality forecasting systems are running routinely in Europe. Of these the ensemble forecasts brought together within the MACC project (http://gems.ecmwf.int/d/products/raq/) represent the state of the art in regional air quality forecasting. On the city level there are also a number of air quality forecasting systems.
Significant uncertainties remain in some aspects of air quality modelling, especially in input data such as emissions and the meteorological fields used for forecasting purposes. One method for obtaining an estimate of uncertainty is through ensemble forecasting, using either an ensemble of models or a single model ensemble. However, apart from the regional-scale ensemble forecasts from MACC, there are currently no operational air quality forecast systems which provide any uncertainty assessments with their forecasts.
Current trends in air quality forecasting will see an increase in both the application of data assimilation techniques, where monitoring and modelling are combined, and the use of ensemble prediction methods. The specific question of uncertainty has only recently begun to be addressed in both assessments and forecasts and this trend is set to continue. Most recent developments have focussed on the regional scale, but these do not address the high pollution episodes that are of most concern in city environments.
In UncertWeb, existing air quality and meteorological models will be applied to produce an ensemble forecast at the urban scale, providing probabilistic forecasts of urban air quality for the city of Oslo. The Ensembles will be based on the ensemble meteorological forecasts from ECMWF, the ensemble of MACC regional air quality models, uncertainties in emissions and uncertainties in model formulation. These probabilistic forecasts of air quality will provide a use case for the web-based developments in UncertWeb.